Project Summary Tongue cancer and the treatments used to cure the disease such as glossectomy can have debilitating effects on speech and swallowing, yet not much is known about the relationship between tongue movements in speech and reconstruction choices after cancer resection. The tongue?s vital role in speaking and swallowing is executed by a uniquely complex array of muscles which deform locally into functional units. Tongue movements are governed by synergistic activation of these functional units using orthogonally oriented muscle fibers not a rigid skeletal structure. Functional units, which are regions of 3D coherent motion, are the structures that link low-level muscle activity to high-level surface tongue geometry. This link, the organization of functional units, can provide considerable insight into normal motor control and adapted tongue motion, as in patients who have had surgery for tongue cancer. However, there is poor understanding of the details of how muscle morphology contributes to function due to tongue muscle interdigitation and complex local activation. A key to understanding the relationship between the structure and function of the tongue is to identify functional units of motion in localized tongue regions and map them to muscle anatomy. A large step in this direction can be gained by developing a tool to cluster 3D coherent motion patterns from tagged MRI, and register them to muscle anatomy from high-resolution MRI. Speech motor control is a critical area of importance where platform tools for machine learning techniques are lacking. The goal of this project is to create a framework to relate the functional units to tongue muscle anatomy. This framework will enhance tongue motion analysis in medical research and clinical applications, where MRI is frequently used. To address this goal, we have already developed technologies to analyze high-resolution and tagged-MRI. The first is a detailed 3D muscle atlas based on 3D high-resolution MRI, which depicts the locations of both extrinsic and intrinsic muscles, and will be registered to each speaker to identify and track their muscles in the 3D tagged data sets. The second is software to derive 3D displacement and strain fields from tagged MR images. These displacements, strains, and derived quantities will be used to extract spatio-temporal feature vectors that define coherent regions, which will be verified with simulation. The vectors will then be input into a novel computational method using a NMF and clustering, the goal of which is to establish the functional units of speech. In addition, biomechanical simulations will be used to co-validate our findings. The concept of functional units is at the core of the proposed work, as it can provide insights into tongue muscle coordination in patients after glossectomy surgery. The research plan comprises two specific aims: (1) develop a computational method to establish functional units from MRI and (2) determine the relationship between alterations in anatomy and alterations in functional unit motion after tongue cancer treatment. Linking the functional units to the anatomy will allow the design of more focused treatments and therapies for tongue motion-related disorders.